Feedback Trading and Its Implications for Return Autocorrelations in India During COVID

Abstract

In emerging capital markets, feedback trading is a widely pursued strategy by investors. Such behavior may potentially lead to volatility and cause negative autocorrelation in market returns, especially during high volatility. In India, such a linkage has not been explored so far, though institutional investors have pursued feedback trading for the last two decades. Also, COVID-19 has led to higher volatility in the markets and might have altered investors’ behavior. This article focuses on finding whether feedback trading is still pursued by institutional investors in Indian equity markets post-COVID and also whether the presence of feedback traders exerts any influence on autocorrelations in market returns. Asymmetric GARCH models are employed to explore the linkage. Findings suggest that while foreign institutional investors continue to pursue positive feedback trading, as in the pre-pandemic period, domestic investors pursue negative feedback trading. However, in the post-pandemic period, as other types of trading became weak or perished, positive feedback traders have started dominating, leading to negative autocorrelation in market returns during heightened volatility. Evidence of negative autocorrelation was not present in the pre-pandemic period. Further, negative news leads to more volatility in returns.

JEL Codes: F21, F32, G11

Keywords

Autocorrelation institutional investors trading behavior post-COVID volatility

1. Introduction

Feedback trading is one of the widely used trading strategies pursued by institutional investors in several advanced and emerging markets. Positive feedback trading (PFT) is a kind of return-chasing behavior pursued by investor groups such as mutual funds and insurance companies, among others. Two important forms of investors or traders in the stock markets whose trading behaviors have been captured by Shiller (1984), Sentana and Wadhwani (1992), Cutler et al. (1990), and Malyar (2005), among many others, are: (a) feedback traders whose demand is based on the analysis of historical indexes and who react based on stock price changes; and (b) smart money traders (or rational investors) who base their asset decisions on market fundamentals and react rationally to anticipated stock returns subject to their wealth limitation. If traders strategize on purchasing stocks during their price increases (decreases) and selling those stocks during decreases (increases) in prices, they are called positive (negative) feedback traders. With stock price decline (rise), an investor’s risk aversion increases (decreases) as his wealth tends to decrease (increase). As investors become exceedingly risk-averse, it leads to more PFT as portfolio protection is sought (Malyar, 2005). Similarly, Sentana and Wadhwani (1992) observe that as smart money investors are risk averse, higher expected volatility encourages them to hold less stock. Feedback trading, if pursued by a group of traders, is associated with autocorrelation in equity markets and mispricing. Further, negative feedback trading dominates when volatility is low, and PFT dominates at high levels of volatility. Thus, the interaction between rational and feedback traders is likely to produce first-order autocorrelation in market returns, as the present return exhibits some relation to past returns due to the presence of feedback traders.

Feedback trading is observed across all asset classes, types of investors, and markets over time (Economou et al., 2022). However, institutional investors are found to engage in feedback trading markedly in domestic stocks with small capitalizations and more prominently in emerging or frontier markets. Further, institutional investors engage in PFT more than individual investors (Nofsinger & Sias, 2002). In the US market, Lakonishok et al. (1992) find little evidence of PFT by institutional investors in the largest stocks and somewhat stronger PFT limited to smaller stocks. These suggest that institutional investors¹ do not essentially or necessarily destabilize stock prices.² However, in the presence of positive feedback from traders, purchases by rational speculators can stimulate PFT by other investors, which may destabilize the market (De Long et al., 1990). Evidence of PFT in emerging capital markets is typically observed in declining markets; this occurs at the time when autocorrelations of stock returns become negative and there is increased volatility (Koutmos & Saidi, 2001). So, feedback traders can incite stock prices to deviate from their intrinsic values, encourage serial correlation in stock returns, and magnify risk in emerging markets. Shiller (1989) argues that PFT may bring about negligible, even negative, autocorrelation.

Empirical evidence indicates that group behavior and feedback trading are more pronounced in developing countries vis-à-vis markets in developed countries (Bohl & Siklos, 2004). While developed market investors appear to follow fundamental-driven investment strategies, emerging market investors appear to be feedback traders; such traders pursue information-guided trading influenced by arbitrage opportunities arising from price deviations between ETF shares and the net asset values of their underlying assets (Da Costa Neto et al., 2019).³

Autocorrelations in market returns may be caused not only by PFT but also by non-synchronous trading⁴ or by the fact that expected returns on stocks share a common, positively autocorrelated process.⁵ To find out whether PFT leads to autocorrelation, it is to be tested by investigating return characteristics during periods of high volatility. Several studies have explored whether market autocorrelation is attributed to PFT in the context of emerging economies, and the evidence is mixed. India is an emerging market where the presence of feedback trading by foreign institutional investors (FIIs) is confirmed by several studies (e.g., Arora, 2016; Chakrabarti, 2001; Choudhary et al., 2022; Gordon & Gupta, 2003; Mukherjee & Roy, 2011, among others). Studies on India majorly focus on whether such trading is pursued by foreign and/or domestic institutional investors (DIIs). Most of the studies employ Granger causality or vector autoregression. However, the linkage between feedback trading and the presence of autocorrelation in market returns has not been studied for the Indian market, though it is well explored in several other emerging economies. Also, most studies on emerging markets have applications of the Sentana and Wadhwani (1992) model with volatility estimated by the EGARCH model, with very rare exceptions that use other variants of the GARCH model.

India is an interesting case, as pieces of evidence show that FIIs pursue PFT, but DIIs pursue negative feedback trading. Also, FIIs play an important role in the Indian stock markets. According to the Global Financial Stability Report of the International Monetary Fund (2020), international investment in equity as a percentage of market capitalization in Q2 of 2019 was around 10 percent for each of China, India, and Chile.

Here lies a research gap that the present article tries to address. It is worth exploring whether engaging in feedback trading has led to higher autocorrelations in India for three reasons: (a) India is an emerging market where feedback trading is pursued for many years; (b) COVID-19 has led to higher volatility in the markets, which might have altered their trading behavior and so examining return autocorrelations during heightened volatility will be insightful for market participants; (c) country experiences are not the same regarding PFT, and analysis concerning India may provide fresh insights.

The present study explores the nature of the feedback trading behavior of institutional investors in India. It also looks at whether the behavior has changed after the COVID pandemic and whether the presence of positive feedback traders leads to negative autocorrelation in stock market returns with heightened volatility during the post- and pre-COVID periods. While the nature of feedback trading is examined by estimating regression and the Granger causality test, for autocorrelation, the methodology suggested by Sentana and Wadhwani (1992) is applied. However, along with exponential GARCH, volatility is also estimated by the GJR-GARCH model, which adds more credence and robustness to the results. This article contributes to the existing literature not only by addressing the research gap mentioned above but also by focusing on the comparison of the linkage during the pre- and post-COVID periods. It is a comprehensive study, as both benchmark stock indexes are considered.

The structure of the remaining article is as follows: the next section presents a brief account of the literature on institutional investors’ trading behavior in emerging markets, including India, and the causes and effects of such trading strategies. Section 3 incorporates the data and methodology used, followed by Section 4, which presents the results and discussion related to them. Section 5 concludes the article.

2. Review of Literature

The existing literature on trading strategies pursued by investors is well-explored and still growing. This section presents the literature on feedback trading in emerging markets and is divided into two subsections: the first one collates the works on the key aspects of PFT, the implications of such trading strategies for volatility, the linkage between feedback trading and market return autocorrelation, and also the impact of any shock or crisis on such trading behavior; the second subsection presents the findings related to feedback trading in the Indian equity market specifically.

2.1 Positive Feedback Trading in Emerging Markets

One strand of literature on PFT focuses on finding out the existence of such trading and its effects on capital market returns and volatility. The existence of PFT is observed in the majority of emerging markets. For instance, Malyar (2005), based on the major stock market indexes of 12 transition economies for the period 1994–2005, observed the presence of PFT in 8 out of 12 economies. Interestingly, in Malaysian stock markets, while FIIs are seen to act as momentum traders, foreign retail investors act as contrarian traders (Sapian & Auzairy, 2015). In China, positive feedback traders are observed to be more inclined to trade when their sentiment is either relatively high or relatively low and tend to choose the moment when most securities move together (Dai & Yang, 2018). Further, some important deductions about the influence of institutional investors in the Chinese securities market (Hong, 2011) were: (a) In a market where PFT occurs with great intensity and is common, institutional investors fail in stabilizing the scenario and tend to follow the trend to amplify such trading and bring about more securities price fluctuations; (b) In a market where such momentum trading is not overriding, institutional investors’ rational arbitrage is not advantageous at all times; on the other hand, irrational investors following PFT do not always face losses and may endure for a long term. Mukherjee et al. (2002) find that FIIs in India pursue feedback trading, and market returns are not influenced by their investments. Other studies find the existence of momentum or PFT by FIIs in emerging markets (e.g., Chauhan & Chaklader, 2020; Mukherjee & Tiwari, 2022, in India; Richards, 2004, in six Asian emerging markets; Choe et al., 1999, in Korea, etc.).

The evidence in the second strand of literature is mixed when it comes to the effect of such trading on the volatility of returns. Institutional short-sellers pursuing PFT are often blamed for intensifying stock market downturns, and for this concern, regulators tend to ban short sales to stabilize stock markets (Bohl et al., 2013). By comparing certain restricted financial stocks with a matched group of unrestricted ones in the United States, UK, Germany, France, Australia, and South Korea during the global financial crisis, Bohl et al. (2013) find that short-sale constraints essentially magnify PFT, thereby destabilizing markets. However, Antoniou et al. (2005) investigated stock price indexes and their corresponding index futures in Canada, France, Germany, Japan, the UK, and the United States and inferred that futures trading lessens the effect of PFT in the underlying spot markets, thus helping to stabilize prices.

The third strand of literature explores the linkage between the existence of feedback trading and market return autocorrelations, particularly during high volatility. This is tested following Sentana and Wadhwani (1992) in the majority of the studies. For example, Watanabe (2002) articulates on the Tokyo Stock Exchange that margin trading causes a substantial amount of PFT and also substantiates the results that stock returns display positive (negative) autocorrelation when volatility is low (high), as observed by Sentana and Wadhwani (1992) and Koutmos (1997) earlier. A similar result is also observed by Bohl and Reitz (2002): a substantial percentage of investors in the German stock market pursue PFT, which induces negative return autocorrelation when periods are characterized by high volatility. However, Angelovska (2013) observes that in small emerging stock markets, high volatility is followed by positive autocorrelation and a positive feedback strategy. Based on the premise that trading volumes are perceived to provide vital hints to forecast future stock price movements, Miwa and Ueda (2011) infer that in the US market, volume-return relations can be attributed to not only the gradual incorporation of fundamental information but also short-term PFT by investors. They observe that autocorrelations of stock returns magnified following large volume rises during the period 1987–2006 and demonstrate how this could be attributed to PFT.

In several Muslim emerging markets, too, from 2001 to 2016, feedback trading was significant in several of these markets, particularly outside Ramadan and not within Ramadan, and this fact remained the same before and after the global financial crisis, implying the possibility that the widely documented lower volatility associated with Ramadan is related to the reduced extent of feedback trading during that particular month (Andrikopoulos et al., 2020).

Another strand of literature includes studies on the impact of a crisis or a shock in the past on the behavior of PFT. One of the impacts can be a rise in PFT with heightened volatility during an ongoing crisis. In some emerging markets in Asia, evidence of positive feedback was observed from January 1996 to October 2004, with a stronger presence in nations characterized by more volatility or that suffer more during a crisis or during and post a crisis (Hsieh et al., 2011). In the Indian equity market, pursuance of negative (positive) feedback trading by FIIs before (during) the global financial crisis was observed (Srinivasan & Kalaivani, 2010). In the Korean market before and during the currency crisis, non-resident institutional investors were more likely to pursue positive-feedback trading in comparison to their resident counterparts (Kim & Wei, 2002). Similarly, Choe et al. (1999) observed a strong indication of PFT by foreign investors before Korea’s economic crisis, while such behavior almost disappeared during the crisis period.

2.2 Feedback Trading in the Indian Equity Market

In the Indian equity market, institutional investors include two prominent groups, viz., FIIs, which include foreign mutual funds, pension funds, etc., and DIIs, which include domestic mutual funds, banks, insurance companies, and development finance institutions. The role of these institutional investors in the Indian stock market is quite well explored. Post-liberalization in the 1990s, along with FIIs investing in India and other emerging markets, DIIs have also been playing a significant role in providing liquidity to the market.

Foreign investors were allowed to invest in the Indian equity market only in the 1990s. Some of the earlier studies observed that FIIs pursue PFT when they invest in equity. For instance, Mukherjee et al. (2002), based on daily data from January 1999 to May 2002, observe that FII flows are significantly impacted by domestic equity market returns, implying return-chasing behavior (see also Chakrabarti, 2001; Gordon & Gupta, 2003). They also observe that factors affecting the purchase and sale decisions of FIIs are not the same. Choudhary et al. (2022) find evidence of FIIs in India following PFT on short horizons.

DIIs in the Indian context, on the other hand, are found to be contrarian or negative feedback traders (Arora, 2016; Kadanda & Raj, 2017; Sathish, 2020). So, FIIs and DIIs exhibit opposite trading behavior. From 2000 to 2006, while FIIs pursued PFT, domestic mutual funds did not exhibit any such behavior as they did not track equity returns (Mukherjee & Roy, 2011). Arora (2016), investigating the period 2007–2013, also found evidence of opposite trading patterns, that is, FIIs and DIIs pursue positive and negative feedback trading, respectively (see also Kadanda & Raj, 2017; Chauhan & Chaklader, 2020).

Regarding the effect of PFT on volatility, evidence varies from time to time. For example, Batra (2003) observes that the trading behavior of FIIs is not destabilizing. Chauhan and Chaklader (2020) find that because of trend-chasing by the FIIs, stock prices move away from their fundamental levels, thereby increasing volatility in index returns. On the other hand, DIIs using the strategy of negative feedback trading provide stability to stock returns. Naik et al. (2021), in a recent study on institutional activities in India amidst the novel COVID-19 pandemic, observed that the spread of the pandemic did not significantly impact stock market volatility, and foreign portfolio investors’ momentum buys and contrarian sales bring about market volatility, whereas DII’s (viz., domestic mutual funds’) trading style does not significantly impact volatility.

3. Data and Methodology

3.1 Data

To find out whether feedback trading persists in the Indian equity market after the COVID-19 pandemic has set in, we adopt two approaches. First, since evidence points out that institutional investors used to pursue such strategies in the pre-COVID period, we examine whether both FIIs and DIIs are adopting this strategy during the COVID period. We also look at whether they pursue a positive or negative feedback trading strategy. Second, we try to examine whether the existence of a large number of positive feedback traders leads to negative autocorrelations in stock market returns pre- and post-COVID periods.

We take the daily data on net investments of DIIs in both the National Stock Exchange (NSE) and Bombay Stock Exchange (BSE) together, denoted as DIIN, and the net investment in equity and debt separately of FIIs on both the stock exchanges together, denoted as FIIEN and FIIDN, respectively. Net investment refers to the difference between the purchase and sale of the concerned asset. The trend in such investments by DIIs and FIIs during the sample period is presented in Figures 1 and 2. It may be noted from Figure 1 that volatility seems to be higher in equity investments by FIIs than in their debt investments. Also, volatility in equity investment has increased in the post-pandemic period compared to the pre-pandemic period. Similarly, from Figure 2, it is evident that volatility in investments by DIIs is also higher during the pandemic compared to the pre-pandemic period. The increase in volatility during the pandemic is also evident from Table 1, in the higher standard deviations in FII investments and DII investments, as well as in stock indexes. Also, for in-depth analysis, the data on purchases and sales are taken separately.

Figure 1.

Source: NSDL.

Figure 2.

Source: Moneycontrol.com.

Table 1.

Standard Deviation in Investments and Stock Prices During the Pre- and Post-COVID Times.

	DIIN	FIIEN	FIIDN
Pre-COVID	791.16	1,426.50	1,277.46
Post-COVID	1,613.10	2,882.27	1,605.47
	Sensex	BSE200	NIFTY
Pre-COVID	5,105.15	661.89	1,493.18
Post-COVID	6,595.93	840.59	1,953.91
	Lsensex*	LBSE200*	Lnifty*
Pre-COVID	0.16189	0.16770	0.15814
Post-COVID	0.16582	0.16944	0.16705
*Logarithm.

The data ranges from April 1, 2014, to March 31, 2021. Since the study intends to find out the impact of the coronavirus pandemic on the feedback trading patterns of institutional investors, we propose to choose a sample period after the effect of the financial crisis of 2007–2008 was over. Taking cue from the increasing trend in GDP growth and surge in the stock markets, we have taken the daily data from April 1, 2014, onward. To find out whether feedback trading behavior has changed with the pandemic compared to the pre-COVID period, the sample is split into pre- and post-COVID periods. January 31, 2020, was the day the first case of COVID-19 was officially announced in India. Hence, from that day onward, the sample period until March 31, 2021, is considered the post-COVID period. The data leads to 1,671 observations in all, out of which there are 283 observations post-COVID.

The data on DIIs is sourced from moneycontrol.com, and the data on FIIs is sourced from the National Securities Depository Limited (NSDL) website. To find out whether and how PFT is affecting the market return and volatility, as proposed by Sentana and Wadhwani (1992), stock market indexes are taken. Benchmark indexes for the two key stock markets in India, viz., NSE and BSE, are taken for the analysis, that is, Nifty 50 for NSE, denoted as NIFTY, and BSE Sensex, denoted by SEN, as well as BSE 200 (denoted by BSE) for BSE (Figure 3). Sensex consists of the 30 largest, most liquid, and financially sound companies’ stocks across key sectors, and we additionally consider a broad-based index, the BSE 200. Nifty 50 is a diversified index with 50 stocks from 13 sectors, and as of March 29, 2019, it represents about 66.8 percent of the free-float market capitalization of the stocks listed on NSE.⁶ The data on NIFTY is sourced from the Yahoo Finance website, while the Sensex and BSE 200 data are sourced from the BSE Ltd website.

Figure 3.

Source: BSE website and Yahoo Finance website.

3.2 Methodology

First, all the variables are tested for unit root, and the results of the augmented Dickey–Fuller test indicate that each of the series DIIN, FIIEN, and FIIDN is stationary. As expected, NIFTY, SEN, and BSE, taken in logarithmic form, are integrated of order one at level, and their first difference series, that is, the return series, are stationary. The return series are prefixed by “R” with their respective names.

To verify the nature of feedback trading pursued by institutional investors (i.e., positive or negative), regression equations of FII equity and debt investment and DII net investment on BSE returns are estimated. Then, to examine whether both FIIs and DIIs are adopting feedback trading in the pre- and post-COVID periods, the Granger causality test of up to five lags is applied for pre- and post-COVID periods separately.⁷

To address the second question on whether and how the existence of PFT is affecting the aggregate market return and volatility, we estimate the Sentana-Wadhwani model (see Sentana & Wadhwani, 1992, for details) for each of the three indexes, viz., NIFTY, SEN, and BSE, and it is estimated for pre- and post-COVID periods, separately.

Following is a brief description of the Sentana–Wadhwani model.

In a market, there are two types of investors: feedback traders and smart money investors. The presence of both in a market forms the theoretical basis for autocorrelated returns and the role of volatility in such autocorrelations. Demand for stocks by feedback traders is γR_t _{– 1}, where R_t _{– 1} denotes past returns, γ > 0 for positive feedback traders and γ < 0 for negative feedback traders. Demand for stocks by smart money traders is the ratio of expected excess return (i.e., return over risk-free return α₀) and the riskiness. The riskiness is modeled as a positive function of conditional variance $σ_{t}^{2}$ . If only smart money traders were there, then $E_{t - 1} R_{t} - α_{0} = α_{1} σ_{t}^{2} .$ But if both types of traders are there, then

\frac{E_{t - 1} R_{t} - α_{0}}{α_{1} σ_{t}^{2}} + γ R_{t - 1} = 1

(1)

If there is a rational expectation, that is, R_t = E_t _{– 1} R_t + ε_t, then the autocorrelation becomes important. The equation becomes:

R_{t} = α_{0} + α_{1} σ_{t}^{2} - (α_{1} γ σ_{t}^{2}) R_{t - 1} + ε_{t}

(2)

If there is PFT, then this will lead to negative autocorrelation, and if there is negative feedback trading, there will be positive autocorrelation.

Now, in addition to these traders, there may be non-synchronous trading or the non-trading problem that induces a positive serial correlation in returns. For instance, a less frequently traded security may respond to news more slowly than a more frequently traded security. So the response of less frequently traded securities with a lag will lead to a positive autocorrelation between the returns of these two securities. This may be reflected in the market index if such non-synchronized trading in the market is common. Hence, in Equation (2), there will be an additional term involving R_t_{− 1}.

In line with the model described above, as per Shiller (1984) and Sentana and Wadhwani (1992), the equation that we estimate is:

R_{t} = α_{0} + α_{1} σ_{t}^{2} + (γ_{0} + γ_{1} σ_{t}^{2}) R_{t - 1} + ε_{t}

(3)

where R_t and R_t_{− 1} are the returns in periods t and t – 1, respectively; R_t is calculated as (ln P_t − ln P_t_{− 1}); α₀ is the rate of return on a risk-free asset, that is, measure of return when rational investors don’t hold shares, $σ_{t}^{2}$ is the conditional variance of returns while α₁ measures the impact of rational investors on returns; it will be negative when the expected volatility rises as rational investors influence prices negatively. γ₀ is the rate of return autocorrelation from non-synchronous trading or negative feedback trading, which causes positive autocorrelation in the returns. At low levels of volatility, negative feedback trading and/or non-synchronous trading dominate, implying positive autocorrelation, and at high levels of volatility, PFT dominates. With the growing risk level, the impact of positive feedback traders γ₁ grows and may indicate a negative autocorrelation due to their dominance. Thus, the interaction between intelligent and positive feedback traders may lead to negative autocorrelation in periods of high volatility.

The conditional variance is modeled as an asymmetric GARCH model, viz., Exponential GARCH (EGARCH), proposed by Nelson (1991), where log-transformed conditional variance is taken. Equation (1) is jointly estimated with the following:

\ln (σ_{t}^{2}) = ω + β_{0} | \frac{ε_{t - 1}}{σ_{t - 1}} | + β_{1} (\frac{ε_{t - 1}}{σ_{t - 1}}) + β_{2} ln (σ_{t - 1}^{2})

(4)

Where ω is the long-term average volatility, $| \frac{ε_{t - 1}}{σ_{t - 1}} |$ is the absolute value of innovations, and $(\frac{ε_{t - 1}}{σ_{t - 1}})$ is innovation. Statistical significance of β₁ implies asymmetry, that is, negative sign implies that volatility increases more when there is bad news. This relates to the leverage effect, which implies that a fall in stock prices raises the debt-to-equity ratio. In this situation, investors want to have higher returns, and stock return volatility increases.

Equations (3) and (4) are estimated jointly, where the initial values for the GARCH parameter in the mean equation 3 are obtained using the OLS method first. To check the robustness of the results, we have also applied the GJR-GARCH model as per Glosten et al. (1993) and compared the autocorrelations obtained therefrom.

4. Results and Discussion

The results of the regression are presented in Table 2. It is observed that while the coefficients of BSE return for up to three past days are positively significant for FII equity and debt investment, they are negatively significant for DIIs. This implies that FIIs are engaged in PFT for the entire sample period, while DIIs are mainly negative feedback traders. This is in line with the findings by Kadanda and Raj (2017), Chauhan and Chaklader (2020), and Mukherjee and Tiwari (2022), among others.

Table 2.

Regression of FIIs and DIIs’ on Stock Return.

Dependent Variable: FIIEN		Dependent Variable: FIIDN		Dependent Variable: DIIN
Variable	Coefficient	Variable	Coefficient	Variable	Coefficient
C	60.04	s.e.	73.76*	C	132.69***
s.e.	49.32		38.68	s.e.	25.92
RBSE_{t – 1}	23974.13***	RBSE_{t – 1}	6,715.87**	RBSE_{t – 1}	–17,108.16***
s.e.	4233.845	s.e.	3,298.17	s.e.	2,054.35
RBSE_{t – 2}	20968.7***	RBSE_{t – 2}	9,215.72***	RBSE_{t – 2}	–15,538.96***
s.e.	4600.75	s.e.	3,558.57	s.e.	2,303.89
RBSE_{t – 3}	13616.77***	RBSE_{t – 3}	7,261.62**	RBSE_{t – 3}	–8,386.398***
s.e.	4560.59	s.e.	3,550.52	s.e.	2,329.52
FIIEN_{t – 1}	0.188976***	FIIDN_{t – 1}	0.25***	DIIN_{t – 1}	0.16***
s.e.	0.039	s.e.	0.034	s.e.	0.034
FIIEN_{t – 4}	0.149864***	FIIDN_{t – 2}	0.13***	DIIN_{t – 2}	0.26***
s.e.	0.026	s.e.	0.027	s.e.	0.032
DCOVID	187.97	FIIDN_{t – 3}	0.13***	DIIN_{t – 3}	0.15***
s.e.	117.7152	s.e.	0.027236	s.e.	0.031
DCOVID1*FIIEN_{t – 1}	–0.033	DCOVID	–184.24**	DCOVID	–19.005
s.e.	0.056	s.e.	93.035	s.e.	58.84
		DCOVID1*FIIDN_{t – 1}	0.029	DCOVID1*DIIN_{t – 1}	0.029
		s.e.	0.058	s.e.	0.051
Adjusted R-squared	0.14	Adjusted R-squared	0.22	Adjusted R-squared	0.36
Durbin–Watson stat	2.1	Durbin–Watson stat	2.04	Durbin–Watson stat	1.996

Note: *** 1% los; ** 5% los; *10% los.

The results of the Granger causality test for FIIEN, FIIDN, and DIIN with respect to Sensex, BSE, and NIFTY are presented in Table 3. The F-statistic for the null hypothesis that market return (for each of Sensex, BSE200, and NIFTY) does not influence DII investment and is rejected at the one percent level of significance both in pre- and post-pandemic periods implying the existence of feedback trading behavior on their part. However, interestingly, the causality also runs from DIIN to market returns (each of the three indexes) in the post-pandemic period, while such influence was not there during the pre-pandemic period. Also, evidence of PFT is there both in pre- and post-pandemic periods, when it comes to FIIEN. However, the causality from FIIEN to returns is not prominent in any of the two periods. Feedback trading was not present in the case of debt investment, that is, FIIDN, in the pre-pandemic period, but some evidence is there for PFT when it comes to BSE in the post-pandemic period. However, more importantly, after the pandemic, evidence is there for almost all lags (across all three indexes) that causality runs from FIIDN to returns⁸.

Table 3.

Granger Causality Test for FIIs and DIIs in the Indian Equity Market.

Null Hypothesis	Pre-COVID						Post-COVID
	Lags	1	2	3	4	5	Lags	1	2	3	4	5
	Obs	F-statistic					Obs	F-statistic
RBSE does not Granger Cause DIIN	1171	59.86***	48.37***	39.84***	31.23***	24.01***	245	28.30***	11.1823***	7.39***	7.88***	6.88***
DIIN does not Granger Cause RBSE		1.25	1.85	0.97	0.83	0.91		10.38***	2.66*	0.90	1.05	0.42
RSEN does not Granger Cause DIIN	1068	21.33***	10.87***	5.45***	3.42***	3.27***	218	10.16***	5.05***	1.64	1.21	1.01
DIIN does not Granger Cause RSEN		12.95***	3.79**	2.15*	0.55	1.08		9.15***	6.21***	4.91***	2.83**	3.59***
RNIFTY does not Granger Cause DIIN	1171	52.02***	44.51***	36.87***	29.22***	22.32***	245	28.19***	11.22***	7.34***	7.83***	6.87***
DIIN does not Granger Cause RNIFTY		1.04	1.85	0.86	0.78	0.86		9.78***	2.73245*	0.98	0.88	0.48
RBSE does not Granger Cause FIIDN	1171	2.12	3.35**	1.48	0.83	1.11	245	2.39***	3.18***	3.55***	1.68	3.19***
FIIDN does not Granger Cause RBSE		0.40	0.43	0.38	0.31	0.34		26.41***	12.84***	8.94***	9.21***	4.07***
RSEN does not Granger Cause FIIDN	1068	1.07	0.22	0.39	0.23	0.20	218	5.51***	0.69	0.68	0.43	1.41
FIIDN does not Granger Cause RSEN		2.93	0.58	0.64	0.87	0.88		3.98***	0.64***	0.95***	0.55	1.50
RNIFTY does not Granger Cause FIIDN	1171	2.11	3.53**	1.54	0.87	1.10	245	2.20	2.69	3.57	1.76	3.25***
FIIDN does not Granger Cause RNIFTY		0.35	0.33	0.34	0.24	0.31		26.42***	13.21***	9.13***	8.79***	3.68***
RBSE does not Granger Cause FIIEN	1171	33.48***	22.27***	12.13***	9.19***	14.86***	245	22.80***	20.18***	13.59***	9.63***	5.37***
FIIEN does not Granger Cause RBSE		1.04	1.53	0.90	0.44	0.29		3.87*	1.14	0.55	0.32	0.74
RSEN does not Granger Cause FIIEN	1068	7.97***	4.99***	5.68***	2.58**	2.57**	218	10.59***	3.99***	2.37*	1.76	0.94
FIIEN does not Granger Cause RSEN		3.60*	2.09	1.47	0.70	0.93		3.96***	1.29	1.03	0.85	1.87
RNIFTY does not Granger Cause FIIEN	1171	35***	23.96***	13.71***	9.75***	16.04***	245	22.76***	19.58***	13.37***	9.53***	5.15***
FIIEN does not Granger Cause RNIFTY		1.42	1.88	1.05	0.59	0.37		3.42	1.00	0.52	0.23	0.79

The results of the EGARCH estimation following Equations (1) and (2) for all three indexes are presented in Table 4.⁹ From the mean equation, α₀ is not statistically significant for any of the indexes, but post-pandemic the value is positive whereas pre-pandemic period values are negative, implying that the investors now earn positive (though insignificant) returns even when they do not invest in shares (invests in bonds). α₁ is the GARCH co-efficient, which is positive and significant for NIFTY and BSE during the pre-pandemic period, indicating that investors were compensated by greater returns for taking higher risks. However, in the post-pandemic period, this turns negative and insignificant for all indexes, implying that, on the whole, rational traders don’t essentially influence prices when the volatility changes.

Table 4.

Results of EGARCH Estimation.

	Pre-COVID	Post-COVID	Pre-COVID	Post-COVID	Pre-COVID	Post-COVID
	RNIFTY	RNIFTY	RSEN	RSEN	RBSE	RBSE
	Mean equation
α ₀	−0.0005	0.001	−0.0003	0.001	−0.0006	0.002
α ₁	11.97*	−0.69	9.89	−0.42	13.97**	−1.92
γ ₀	0.095*	0.063	0.09*	0.053	0.14***	0.060
γ ₁	250.48	−97.06**	192.58	−85.76**	73.60	−98.51**
	Variance Equation
ω	−0.77***	−0.29**	−0.66***	−0.30**	−1.05***	−0.27**
β ₀	0.11***	−0.001	0.09***	0.011	0.15***	−0.012
β ₁	−0.17***	−0.21***	−0.15***	−0.21***	−0.18***	−0.21***
β ₂	0.93***	0.97***	0.94***	0.97***	0.90***	0.97***

Note: ***1% los, **5% los, *10% los.

The constant component of the autocorrelation, γ₀, is positive and significant (at a 10 percent level of significance, except BSE at a 1 percent level of significance) during the pre-pandemic period but turned insignificant in the post-pandemic period. This evidence points to the presence of non-synchronous trading in the pre-pandemic period, but the influence of positive autocorrelation ceases to exist after the pandemic. More interestingly, γ₁ is negative and significant at a 5 percent level of significance in the post-pandemic period for all indexes, indicating the presence of PFT leading to negative autocorrelations with higher volatility. Importantly, this coefficient was positive and insignificant during the pre-pandemic period.

Now, it needs to be pointed out that in the previous analysis, we have observed that feedback trading is pursued by both FIIs in equity and DIIs in both the pre- and post-pandemic periods. But their feedback trading behavior did not lead to significant negative autocorrelation in returns in the pre-pandemic period, while such influence is significant in the post-pandemic period. Is it because the size of transactions by FIIs and DIIs has increased over time? The absolute value of net investment in FIIs and DIIs as a proportion of market capitalization at BSE does not show any significant change in pre- and post-pandemic periods (Figure 4). While DII net investment as a proportion of BSE market capitalization was consistently below 1 percent in the entire sample period, barring only 2019–2020, the proportion of FII investment ranged from 0.19 to 2.69 percent in the pre-pandemic period and remained between 0.38 and 1.09 percent in the post-pandemic period. So, despite their size remaining more or less the same compared to the total market size, they have been able to exert an influence on return autocorrelation in the post-pandemic period. This is mainly due to the absence or weaker influence of negative feedback traders and/or the insignificant impact of non-synchronous trading. It is to be noted that such non-synchronous trading used to lead to autocorrelation (positive and significant) in the pre-pandemic period, which seemed to have waned after the pandemic, leading to the negative autocorrelation in returns on account of positive feedback traders during periods of high volatility.

Figure 4.

FII and DII Net Investment as Proportion of BSE Market Capitalisation (%).

In the variance equation in Table 4, almost all the coefficients are statistically significant across all indexes. This implies that current volatility in returns depends on the last period’s squared innovation and the last period’s volatility. β₁ is negative and significant for all market indexes both in the pre- and post-pandemic periods, indicating that the impact of negative news is much larger than the impact of positive news, that is, asymmetry exists; in other terms, this means volatility tends to rise more as a result of negative news. Thus, in both BSE and NSE, the leverage effect is present. Also, the past volatility coefficient, β₂, is positive, highly significant, and close to unity for all indexes both in the pre- and post-pandemic periods, implying a high degree of shock persistence. As β₂ < 1, the stationarity condition for the EGARCH specification holds for all the indexes. In each of the market index specifications, this co-efficient value has increased in the post-pandemic period compared to the pre-pandemic period.¹⁰

We further assess the empirical relevance of PFT by calculating the autocorrelation coefficient ${\hat{ρ}}_{t} = {\hat{γ}}_{0} + {\hat{γ}}_{1} {\hat{σ}}_{t}^{2}$ . We need to check whether negative γ₁ exceeds positive γ₀, leading to negative autocorrelation. In Table 5, we report the autocorrelation coefficient for estimated minimum, mean, median, and maximum conditional volatility; autocorrelations estimated by both EGARCH and GJR-GARCH are presented. The results indicate that in the post-pandemic period, PFT is not prominent at low volatility levels, i.e., at the minimum and mean volatility. However, the effect is evident from the negative and increased value of the autocorrelation coefficient when the volatility is at its maximum. More interestingly, volatility levels have increased in the post-pandemic compared to the pre-pandemic levels, and the positive autocorrelation coefficients have also gone down in the post-pandemic. Also, in the post-pandemic period, with higher volatility, autocorrelations turn out to be negative. This makes us conclude that PFT is a phenomenon that may not be prevalent very widely but is taking place during high volatility. Notably, the autocorrelation patterns are qualitatively similar for both the models applied, which indicates the robustness of the results.

The finding that PFT is pursued by FIIs when they invest in equity is in synchronization with previous studies on India (e.g., Arora, 2016; Gordon & Gupta, 2003;, Kadanda & Raj, 2017; Mukherjee et al., 2002, among others) and the results in Table 2. During the post-pandemic period, too, FIIs continue to exhibit such behavior and this is in tune with Choudhary et al. (2022). But, interestingly, the present study also indicates that FIIs did not track returns for debt investment in the pre-pandemic period, but some evidence in the post-pandemic period indicates they now track market returns. While earlier studies have pointed to mixed evidence regarding DIIs pursuing negative feedback trading (Chauhan & Chaklader, 2020) or no feedback trading at all (Mukherjee & Roy, 2011), the present study finds that DIIs do track market returns and pursue negative feedback trading during the entire sample period.

Table 5.

Volatility and Autocorrelation.

	NIFTY (Pre-pandemic)	NIFTY (Post-pandemic	BSE (Pre-pandemic)	BSE (Post-pandemic)	SENSEX (Pre-pandemic)	SENSEX (Post-pandemic)
EGARCH
Conditional volatility: σ_t²
min	0.00003	0.00008	0.00004	0.00008	0.00003	0.00008
mean	0.00008	0.00039	0.00008	0.00036	0.00008	0.00042
median	0.00007	0.00017	0.00007	0.00016	0.00007	0.00017
max	0.00036	0.00708	0.00053	0.00706	0.00032	0.00756
Autocorrelation co-efficient: ρ
min	0.10	0.05	0.14	0.05	0.10	0.05
mean	0.11	0.03	0.14	0.03	0.11	0.02
median	0.11	0.05	0.14	0.04	0.11	0.04
max	0.18	−0.62	0.18	−0.63	0.16	−0.60
GJR-GARCH
Conditional volatility: σ_t²
min	0.00003	0.00006	0.00003	0.00005	0.00003	0.00006
mean	0.00008	0.00039	0.00008	0.00035	0.00008	0.00041
median	0.00006	0.00015	0.00006	0.00015	0.00006	0.00016
max	0.00078	0.00552	0.00114	0.00506	0.00069	0.00594
Autocorrelation Co-efficient: ρ
min	0.09	0.02	0.13	0.01	0.08	0.01
mean	0.10	0.00001	0.13	−0.01	0.09	−0.01
median	0.10	0.01	0.13	0.001	0.09	0.005
max	0.25	−0.29	0.20	−0.26	0.21	−0.28

Note: ${\hat{ρ}}_{t} = {\hat{γ}}_{0} + {\hat{γ}}_{1} {\hat{σ}}_{t}^{2}$ .

5. Conclusion

This article tries to find out whether institutional investors (both domestic and international) are pursuing feedback trading in the Indian equity market even after the COVID-19 pandemic. Also, it looks into whether the existence of a large number of positive feedback traders leads to negative autocorrelations in stock market returns when volatility increases during the post- and pre-COVID periods following the Sentana–Wadhwani model. Based on daily data from 2014 to 2021, this article employs multiple regression, the Granger causality test, and estimates the Sentana–Wadhwani model for BSE Sensex, BSE 200, and NSE NIFTY.

The findings provide some interesting insights. First, PFT behavior is exhibited by FIIs, while evidence of negative feedback trading is observed to be pursued by DIIs. DIIs, in the post-pandemic period, have become more important in influencing market returns, whereas they did not exert any such influence in the pre-pandemic period. Second, for equity investments by FIIs, PFT is observed both in the pre- and post-pandemic periods. However, for debt, such evidence is rare and, on the contrary, debt investment by them influences market returns in the post-pandemic period. Third, in the post-pandemic period, there is evidence of the presence of PFT leading to negative autocorrelation in the market return when volatility increases, but this was not the case in the pre-pandemic period. This indicates that after the pandemic, positive feedback traders have become more influential in affecting the market autocorrelation, especially because the influence of negative feedback traders and/or non-synchronous trading has ceased to exist. Fourth, there is an asymmetry in the effect of positive and negative shocks on the volatility of returns; volatility is influenced to a greater extent by negative news.

The findings present a transformation in the equity and debt market scenarios after the pandemic. PFT is a phenomenon that may not be prevalent very widely but is taking place during high volatility. A possible extension of this work in the future might be to check that whenever there is an increase in volatility, PFT dominates the market and leads to negative autocorrelation. Another future direction of research may also be exploring the feedback trading behavior of investors in times of various policy changes in different emerging countries, which are expected to have an impact on their decisions.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Notes

References

Andrikopoulos

, Cui

, Gad

, & Kallinterakis

(2020). Feedback trading and the Ramadan effect in frontier markets. Research in International Business and Finance, 51, 101085.

Angelovska

(2013). Detecting positive feedback trading when autocorrelation is positive. Zagreb International Review of Economics & Business, 16(1), 93–101.

Antoniou

, Koutmos

, & Pericli

(2005). Index futures and positive feedback trading: Evidence from major stock exchanges. Journal of Empirical Finance, 12(2), 219–238.

Arora

R. K.

(2016). The relation between investment of domestic and foreign institutional investors and stock returns in India. Global Business Review, 17(3), 654–664.

Batra

(2003). FII trading strategies. ICRIER Working Paper, September.

Bohl

M. T.

, Klein

A. C.

, & Siklos

P. L.

(2013). Are short sellers positive feedback traders? Evidence from the global financial crisis. CIGI papers No. 15, April 2013, 4–17. Retrieved May 18, 2022, from https://www.files.ethz.ch/isn/164550/CIGI_Paper_15.pdf

Bohl

M. T.

, & Reitz

(2002). The influence of positive feedback trading on return autocorrelation: Evidence for the German stock market. Working Paper Series of Postgraduate Research Programme “Capital Markets and Finance in the Enlarged Europe.”

Bohl

M. T.

, & Siklos

P. L.

(2004). Empirical evidence on feedback trading in mature and emerging stock markets. Research Paper Series 137, Quantitative Finance Research Centre, University of Technology, Sydney.

Chakrabarti

(2001). FII flows to India: Nature and causes. Money and Finance, 2(7), 61–81.

10.

Chauhan

A. K.

, & Chaklader

(2020). Do local investors exhibit smart value investment? Empirical evidence from India. Global Business Review. 10.1177/0972150920915330

11.

Choe

, Kho

B-C.

, & Stulz

R. M.

(1999). Do foreign investors destabilize stock markets? The Korean experience in 1997. Journal of Financial Economics, 54(2), 227–264.

12.

Choudhary

, Singh

, & Soni

(2022). Relationship between FIIs’ herding and returns in the Indian equity market: Further empirical evidence. Global Business Review, 23(1), 137–155.

13.

Cutler

D. M.

, Poterba

J. M.

, & Summers

L. H.

(1990). Speculative dynamics and the role of feedback traders. American Economic Review, 80(2), 63–68.

14.

da Costa Neto

A. F.

, Klotzle

M. C.

, & Pinto

A. C. F.

(2019). Investor behavior in ETF markets: A comparative study between the US and emerging markets. International Journal of Emerging Markets, 14(5), 944–966.

15.

Dai

Z-M.

, & Yang

D-C.

(2018). Positive feedback trading and investor sentiment. Emerging Markets Finance and Trade, 54, 2400–2408.

16.

De Long

J. B.

, Shleifer

, Summers

L. H.

, & Waldmann

R. J.

(1990). Positive feedback investment strategies and destabilizing rational speculation. Journal of Finance, 45(2), 379–395.

17.

Economou

, Gavriilidis

, Gebka

, & Kallinterakis

(2022). Feedback trading: A review of theory and empirical evidence. Review of Behavioural Finance. 10.1108/rbf-12-2021-0268

18.

Glosten

L. R.

, Jagannathan

, & Runkle

D. E.

(1993). On the relation between the expected value and the volatility of nominal excess return on stocks. Journal of Finance, 48, 1779–1801.

19.

Gordon

J. P. F.

, & Gupta

(2003). Portfolio flows into India. IMF Working Papers, (020), A001. Retrieved May 29, 2022, from https://www.elibrary.imf.org/view/journals/001/2003/020/article-A001-en.xml

20.

Hong

(2011). Positive feedback trading, institutional investors and securities price fluctuation. China Finance Review International, 1(2), 120–132.

21.

Hsieh

M-F.

, Yang

T-Y.

, Yang

Y-T.

, & Lee

J-S.

(2011). Evidence of herding and positive feedback trading for mutual funds in emerging Asian countries. Quantitative Finance, 11(3), 423–435.

22.

International Monetary Fund. (2020). The global financial stability report, April 14.

23.

Kadanda

, & Raj

(2017). Relationship between foreign portfolio investments (FPI), domestic institutional investors, and stock market returns in India. International Journal of Financial Management, 7(4), 1–9.

24.

Kim

, & Wei

S-J.

(2002). Foreign portfolio investors before and during a crisis. Journal of International Economics, 56, 77–96.

25.

Koutmos

(1997). Feedback trading and the autocorrelation pattern of stock returns: Further empirical evidence. Journal of International Money and Finance, 16(4), 625–636.

26.

Koutmos

, & Saidi

(2001). Positive feedback trading in emerging capital markets. Applied Financial Economics, 11, 291–297.

27.

Lakonishok

, Shleifer

, & Vishny

R. W.

(1992). The impact of institutional trading on stock prices. Journal of Financial Economics, 32, 23–43.

28.

Malyar

(2005). Positive feedback trading and return autocorrelation in the transition markets: Case of CEE countries and selected CIS economies, master’s thesis, National University Kyiv-Mohyla Academy.

29.

Miwa

, & Ueda

(2011). A dynamic volume–return relation and investors’ positive feedback trading. International Review of Finance, 11(3), 325–351.

30.

Mukherjee

, Bose

, & Coondoo

(2002). Foreign institutional investment in the Indian equity market: An analysis of daily flows during January 1999-May 2002. Money & Finance, 2(9–10), 21–51.

31.

Mukherjee

, & Roy

(2011). The nature and determinants of investments by institutional investors in the Indian stock market. Journal of Emerging Market Finance, 10(3), 253–283.

32.

Mukherjee

, & Tiwari

(2022). Trading behaviour of foreign institutional investors: Evidence from Indian stock markets. Asia-Pacific Financial Markets. 10.1007/s10690-022-09361-z

33.

Naik

P. K.

, Shaikh

, & Huynh

T. L. D.

(2021). Institutional investment activities and stock market volatility amid COVID-19 in India. Economic Research-Ekonomska Istraživanja. DOI:10.1080/1331677X.2021.1982399

34.

Nelson

(1991). Conditional heteroskedasticity in stock returns: A new approach, Econometrica, 59, 347–70.

35.

Nofsinger

J. R.

, and Sias

R. W.

(2002). Herding and feedback trading by institutional and individual investors. Journal of Finance, 54(6), 2263–2295.

36.

Richards

(2004). Big fish in small ponds: The trading behavior and price impact of foreign investors in Asian emerging equity markets. Research Discussion Paper 2004-05, Economic Analysis Department, Reserve Bank of Australia. Retrieved from June 1, 2022, http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.323.742&rep=rep1&type=pdf

37.

Sapian

R. Z. Z.

, & Auzairy

N. A.

(2015). Foreign equity flows and market return linkages: Evidence of Malaysian stock market. Global Business Review, 16(5S), 1S–14S.

38.

Sathish

(2020). An analysis of trading behaviour of foreign and domestic institutional investors in the Indian stock market: An empirical study. Indian Journal of Research in Capital Markets, 7(1), 22–37.

39.

Sentana

, & Wadhwani

(1992). Feedback traders and stock return autocorrelations: Evidence from a century of daily data. Economic Journal, 102(411), 415–425.

40.

Shiller

R. J.

(1984). Stock prices and social dynamics. Brookings Papers on Economic Activity, 15(2), 457–510.

41.

Shiller

R. J.

(1989). Market volatility (p. 375). MIT Press.

42.

Srinivasan

, & Kalaivani

(2010). Foreign institutional investment and stock market returns in India: Before and during the global financial crisis. IUP Journal of Behavioral Finance, 7(1–2), 59–75.

43.

Watanabe

(2002). Margin requirements, positive feedback trading, and stock return autocorrelations: The case of Japan. Applied Financial Economics, 12, 395–403.